INDEX

Explanations

abbreviations or phrases like "i.e." within parentheses

oai_token-act-pair · gpt-3.5-turbo

phrases and expressions that involve the word "e"

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rament

-0.65

 Flavoring

-0.63

ONSORED

-0.60

 crunch

-0.60

displayText

-0.59

ingham

-0.59

 seams

-0.58

mable

-0.58

spot

-0.58

 rebate

-0.57

POSITIVE LOGITS

lements

0.85

clusive

0.83

raq

0.82

lez

0.82

lege

0.82

clair

0.81

tainment

0.80

gypt

0.78

.,

0.78

astern

0.76

Activations Density 0.018%

abbreviations or phrases like "i.e." within parentheses

phrases and expressions that involve the word "e"

No Comments

No Known Activations